The safety (and other things) of DIY floating / isolated class 1 PSU's

[Jalca]

Hi all this is my first post so please be gentle. Apologies in advance for any etiquette misdemeanours. I come to electronics through an interest in audio. I’ve been at it for a few years and am perhaps not a beginner but I am inexperienced and an amateur.

I want to better understand the use and safety considerations of DIY isolated / floating class 1 PSU’s, particularly in respect of audio. To be clear a conductive chassis that houses mains leads (active and neutral) are always directly bonded mechanically and electrically to PE. This question relates only to referencing the 0V rail of on the isolated side of a trx to PE.

I’ve been reading and learning about grounding topologies for DIY class 1 linear PSU’s, the circuits they power as used in audio (i.e amps, active xovers, DAC’s etc) and the associated noise that can result. It has become clear that the majority of advice found on forums and websites is that a DIY class 1 isolated / floating PSU’s should not be used i.e. the 0v rail should always be referenced to Protective Earth (PE). A reason is that of safety: should a trx fault occur where mains voltage passes across to the secondary side, the user has a greater chance of protection because fault currents could flow to PE and might trip a protection device.

Examples of this:

1.   Audio Component Grounding and Interconnection by Dave Davenport on diyaudio - https://www.diyaudio.com/archive/articles/186018-what-gain-structure-next-thread.html
2.   Ground Loops - https://hifisonix.com/ground-loops/
3.   Grounding - https://www.valvewizard.co.uk/Grounding.pdf

Another reason given in forum threads is that the use of isolated, floating supplies is problematic for reasons other than safety. Unfortunately these “other (technical?) reasons” are not elaborated upon.

However, there are a small number of people who say that isolated, floating PSU’s are perfectly safe in the audio context and even preferred from a safety perspective. One such person was an old electronics engineer I had a conversation with many years ago who took the position (as I understood it) that once mains voltage had been isolated, it should be left floating / isolated unless there was a reason against this. His reasoning was that an isolated, floating PSU provided a smaller surface area from which electric shock could be obtained, i.e. a risk only existed between 0v rail and the +/-. He argued that referencing to PE increased the surface area that an electric shock could come from, i.e. a person touching the +/- rail and any PE connected (conductive) surface could be shocked. His view was that it could not be relied upon that RCD’s would always be present (an example might be an older home that still relied on CB or fuses). Unfortunately we did not discuss the topic of trx failure and mains appearing on the previously isolated side.

Knowing that lab bench PSU’s (DIY and otherwise) are usually class 1 isolated / floating devices and don’t seem to generate passionate safety arguments I am left confused as to why a floating/isolated PSU in the audio context would be unsafe or be problematic for technical reasons.

Having now come to understand the arguments for always referencing DIY class 1 audio PSU’s to PE and following the grounding topologies in the links shown above to avoid noise, I wish to more fully understand the arguments for why. Vague references to safety and even vaguer references to “other reasons” doesn’t help form a complete understanding of the topic. My goal here is to understand the topic from all sides, enhancing my overall electronics knowledge.

My questions are:


1)   What are the safety concerns in using a floating / isolated class 1 PSU and can they be overcome?
2)   What are the non-safety “technical reasons” that would make using isolated / floating PSU’s problematic in an audio context?
3)   Are those that are using isolated / floating PSU’s in audio exposing themselves and others to unacceptable safety risks as the weight of opinion as per the links above would suggest?


Many thanks to all.

[boB]

I would say that the main concern of a floating supply has to do with the  insulation system in the transformer.

You don't want it to short out when it gets hot or when you tighten it down.

Also be careful of tightening down the bridge rectifier too tight and things like that.

It's not difficult.

boB 🌜

[Ian.M]

+1
Its all down to the transformer, which needs to be designed *AND* constructed with either reinforced (or double) insulation between the primary and secondary or a grounded interwinding screen, and also have a thermally fused primary.  The thermal fuse will blow before overheating can compromise the primary to secondary insulation.

The problem then becomes: Can you trust the transformer manufacturer and the distributor you got it from?   If dealing with 1st world manufacturers and major international distributors, the answer is probably yes.  If dealing with Chinese manufacturers or buying through Ebay, Amazon and far east eBazar sites the answer is generally *NO*.

[Gyro]

As others have said, it relies on you using a decent quality transformer - either a split bobbin standard one (I'm not sure there are many others these days because they are easier to wind too), or a toroidal with an interwinding screen (or reputable manufacturer isolation spec).

Since deep into the mists of time, it has been normal to connect signal ground to PE on only one item in the system (normally the power amplifier as it has the biggest transformer, fuse etc.), in order to avoid hum due to ground loops. It is only relatively recently that Class II equipment has come along, with double insulated metal cases, Y cap leakage currents etc. Whilst these don't normally cause mains hum, you can get very nasty buzzes when inserting phono connectors and occasionally, minor shocks from the skirts caused by Y cap discharges. In extreme cases you can end up with everything in the system Class II, where you get a tingle every time you touch the metalwork, again caused by the current leakage through multiple Y caps.

One technique that used to be used, mainly in stage equipment, guitar amps etc. is the 'ground lift', where a pair or rectifier diodes (1-3A) are connected in inverse parallel between signal ground and the case (PE). The 0.6V before the diodes start conducting is enough to avoid problems with ground loops (which are very low voltage but higher current), whilst still providing some fault protection against signal ground accidentally being connected to something live. This sort of measure normally wouldn't be used in home audio though.

The Main precaution in construction is to physically restrain the mains wiring (by length, cable ties etc). so there is zero possibility of it coming into contact with anything on the secondary / signal ground side of the equipment. This is to cover a mains connection coming loose, solder joint failure etc. EDIT: A particular concern here is normally the longer wiring gtom the rear of the cabinet to a front panel mains switch. This needs to be restrained (P clips etc) and an overall piect of heatshrink covering the switch and extending back well past the solder joints to the first restraining point. Just as important, ensure that secondary side signal wiring is restrained from coming loose and accidentally coming into contact with mains wiring.

[Jalca]

I had suspected that it came down to the trx. It seems it is really a question of design philosophy: rely on trx design and build quality on the one hand or rely on PE, fuses/CB/ RCD’s on the other.

From a safety perspective, is one philosophy generally considered safer than the other or is it simply a matter of pick one and implement it properly?

Gyro:

The ground lift with diodes you describe is the basis of the grounding topologies outlined in the links in my first post, however they specify a 25 – 35A bridge rectifier block in parallel with a 10R resistor and 220nf capacitor, designating it a ‘Ground Loop Breaker’. One of these would then be used in the PSU of each device.

A question on this though: does lifting the ground with just the diodes (leave out the cap and resistor) maintain effective isolation of the secondary windings from PE in terms of a shock hazard? In other words if someone grabbed hold of PE and the +/- rail on a PSU that had its 0V rail connected to PE via diodes would there be a shock hazard? I have measured the dc resistance of some diodes and they are 250k – 300k. Would this be enough of a PE reference?   ****Edit, I think I've answered my own question, the current flow would be tiny. ***

[BrokenYugo]

Diodes are not resistors!!! They are non ohmic devices, so Ohm's law does not apply.

The drop through a rectifier diode is only perhaps a volt, from a safety perspective that would absolutely be a ground bonded secondary.

[Jalca]

Diodes are not resistors!!! They are non ohmic devices, so Ohm's law does not apply.

The drop through a rectifier diode is only perhaps a volt, from a safety perspective that would absolutely be a ground bonded secondary.

Okay, thanks. Yes, I see my error now. I had been thinking about it wrong. Once I drew it out it became obvious. 

[boB]

Yeah, you should not expect very high current on the ground connection unless something is wrong.  Probably that transformer

Leakage current can and does certainly pass that way though.

If you can come by a high-pot or megger HV leakage current tester, you might be more assured of good insulation.

But even then, I would be careful not to adjust the current sensitivity too low (high current trips it) or higher voltage than necessary because too much of that testing can sometimes create a problem where there was not one before.

If it's audio, you can use other isolation techniques as well, like, line level audio transformers if you find good ones at a decent price.  Then there is optical like S/PDIF or something like that.

Even with all those precautions I am always careful not to go adjusting audio levels that are plugged into the mains while standing barefoot in a mud puddle without wearing thick gloves.

boB 🌜

[Gyro]

I had suspected that it came down to the trx. It seems it is really a question of design philosophy: rely on trx design and build quality on the one hand or rely on PE, fuses/CB/ RCD’s on the other.

From a safety perspective, is one philosophy generally considered safer than the other or is it simply a matter of pick one and implement it properly?

Gyro:

The ground lift with diodes you describe is the basis of the grounding topologies outlined in the links in my first post, however they specify a 25 – 35A bridge rectifier block in parallel with a 10R resistor and 220nf capacitor, designating it a ‘Ground Loop Breaker’. One of these would then be used in the PSU of each device.

A question on this though: does lifting the ground with just the diodes (leave out the cap and resistor) maintain effective isolation of the secondary windings from PE in terms of a shock hazard? In other words if someone grabbed hold of PE and the +/- rail on a PSU that had its 0V rail connected to PE via diodes would there be a shock hazard? I have measured the dc resistance of some diodes and they are 250k – 300k. Would this be enough of a PE reference?   ****Edit, I think I've answered my own question, the current flow would be tiny. ***

As I mentioned, the diode ground lift is rarely used. You're not breaking new territory here. Most traditional bench PSUs are Class I devices with PE grounded metal cases and floating output terminals. They often provide a separate ground terminal too, so that you can ground an output if desired, but they are often used floating for test setup reasons (avoiding ground loops again).

As long as you carefully follow the sensible precautions of ensuring that primary side and secondary side wiring cannot mix (and in the case of primary wiring, would hit grounded metalwork before it can get near the secondary side), and use a decent quality transformer then you really don't have a problem. When you get to the construction stage, post pictiures and I'm sure you will receive helpful feedback.

Things get more dodgy with SMPS type power supplies, where you are very dependent on manufacturing quality in deciding whether the secondary side can safely be floated, but from your description, it sounds as if you are talking about a traditional mains frequency transformer linear supply.